An energy efficient clustering and routing protocol with data fault prediction using hybrid EAAM with BiLSTM-GRU-IDS in wireless sensor networks
G. Mahalakshmi, S. Ramalingam, A. Manikandan, S. Murugesan
Abstract
Wireless sensor networks (WSNs) and the Internet of Things (IoT) are essential components in various applications. As WSN nodes often function with limited battery life, optimizing energy efficiency becomes crucial for clustering and routing tasks. Additionally, maintaining the reliability and security of transmitted data presents a significant challenge, especially in environments susceptible to attacks from malicious nodes. This research seeks to overcome these challenges by creating a routing protocol that prioritizes both security and energy efficiency while also integrating fault data prediction to enhance network longevity and data dependability. The process of cluster head (CH) formation and clustering is initially performed using the Enhanced Aphid-Ant Mutualism Optimization (EAAM) method. For reliable path selection, the Bald Eagle Search (BES) algorithm is employed. To further enhance system security and performance, a bidirectional deep short-term memory (BiLSTM) and gated recurrent unit (GRU)-based intrusion detection system (IDS), named Deep BiLSTM-GRU-IDS, is proposed for effective intrusion detection. The goal of the proposed solution is to enhance the accuracy and detection rate of the IDS while reducing processing time, particularly by minimizing the false positive rate in the WSN environment. The performance of the intrusion detection model was assessed using various evaluation parameters on the KDD Cup 1999 dataset, focusing on the detection rate, false alarm rate, and latency rate. The results highlight the effectiveness of the Deep BiLSTM-GRU-IDS, demonstrating its compatibility with other compared algorithms. When compared to existing methods like Multi-Objective Particle Swarm Optimization (MO-PSO), Multi-objective Fractional Particle Lion Algorithm (MOFPL), Firefly Cyclic Randomization (FCR), Adaptive Shark Smell Optimization (ASSO), and Salp Swarm Optimization (SSO), the proposed protocol improves network lifetime by 30.3%, 22.85%, 16.21%, 8.86%, and 7.5%, respectively.